Enigmatic Incongruence Between Mtdna and Ndna Revealed by Multi

Home , Cipangopaludina, Margarya, Semisulcospira morii, Sinotaia

www.nature.com/scientificreports

OPEN Enigmatic incongruence between mtDNA and nDNA revealed by multi-locus phylogenomic analyses Received: 28 August 2018 Accepted: 5 April 2019 in freshwater snails Published: xx xx xxxx Takahiro Hirano1, Takumi Saito 2, Yoshihiro Tsunamoto3, Joichiro Koseki2, Bin Ye2,4, Van Tu Do5, Osamu Miura6, Yoshihisa Suyama3 & Satoshi Chiba2,7

Phylogenetic incongruence has frequently been encountered among diferent molecular markers. Recent progress in molecular phylogenomics has provided detailed and important information for evolutionary biology and taxonomy. Here we focused on the freshwater viviparid snails (Cipangopaludina chinensis chinensis and C. c. laeta) of East Asia. We conducted phylogenetic analyses and divergence time estimation using two mitochondrial markers. We also performed population genetic analyses using genome-wide SNPs. We investigated how and which phylogenetic patterns refect shell morphology. The results showed these two species could be separated into four major mitochondrial clades, whereas the nuclear clusters supported two groups. The phylogenetic patterns of both mtDNA and nDNA largely refected the geographical distribution. Shell morphology refected the phylogenetic clusters based on nDNA. The fndings also showed these two species diversifed in the Pliocene to early Pleistocene era, and occurred introgressive hybridisation. The results also raise the taxonomic issue of the two species.

Molecular phylogeny provides a robust framework for investigations in the felds of taxonomy and conservation biology, such as on species diversity and the patterns of geographical distribution1. For example, molecular phylogenetic studies have clarifed the taxonomic issues of whether species are endemic or alien2 and the existence of cryptic species3. However, by comparing the phylogenetic patterns determined using diferent molecular markers, it has become clear that there is incongruence between the molecular markers even in a lineage determined using the same sample4,5. To resolve this problem, the inference of species trees from multi-locus data can be benef- cial6,7. Although molecular phylogenomics such as a multi-locus approach provides more detailed information than an approach that uses a single or a few genes for applications in evolutionary biology and taxonomy8–10, such study is still limited. Under these circumstances, molluscs are an excellent model to compare the relationships among phylogeny, morphology, and geographical distribution patterns. Tey are the second most diverse phylum of animals11 and molluscs also show a high level of local adaptation and genetic divergence among populations12. In particular, many studies of freshwater snails have documented substantial incongruence between their molecular phylogeny and morphology, suggesting parallel or convergent evolution13,14 along with cryptic speciation15, cryptic inva- sions16, and the retention of ancestral polymorphisms or introgression14. However, there are not many phylogenetic studies of molluscs using genome wide screens with SNPs so far, therefore, studies investigating mechanisms of incongruence using multi-locus data are needed. Here, we focus on the freshwater viviparid snails of Japan. Cipangopaludina chinesis has two subspecies, C. c. chinensis (Gray, 1834) and C. c. laeta (Martens, 1860). Te taxon chinensis is distributed in parts of mainland

1Department of Biological Sciences, University of Idaho, Moscow, Idaho, USA. 2Graduate school of Life Sciences, Tohoku University, Miyagi, Japan. 3Kawatabi Field Science Center, Graduate School of Agricultural Science, Tohoku University, Miyagi, Japan. 4Agricultural Experiment Station, Zhejiang University, Hangzhou, China. 5Institute of Ecology and Biological Resources, Vietnam Academy of Science and Technology, Hanoi, Vietnam. 6Faculty of Agriculture and Marine Science, Kochi University, Kochi, Japan. 7Center for Northeast Asian Studies, Tohoku University, Miyagi, Japan. Correspondence and requests for materials should be addressed to T.H. (email: [email protected])

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 1 www.nature.com/scientificreports/ www.nature.com/scientificreports

Asia such as China and Vietnam, and has not been recorded from Japan. On the other hand, laeta is distributed in Japan, and was listed as vulnerable due to the declining of populations in the Red Data Book 2014, Treatened Wildlife of Japan17. Tese species have also been spread unintentionally through the expansion of agriculture, such as through rice farming18. Terefore, in a previous study it was suggested that C. c. laeta is not native to Japan and was actually introduced from China, as these species are not present in the Pleistocene and Holocene fossil record of Japan18,19. However, in another study, a fossil was potentially identifed as C. c. laeta from a Toyono formation (0.5–0.3 Ma20). In addition, our previous phylogenetic study showed that C. c. laeta of mainland Japan is most likely native because of its genetic diferences from nominotypical subspecies C. c. chinensis according to GenBank data21. However, no comprehensive studies have been performed to clarify the genetic divergence between each population of C. c. laeta in the Japanese Archipelago, including Kyushu and the Ryukyu Islands. In particular, because the Kingdom of the Ryukyus frequently undertook cultural exchange with China and mainland Japan, the population of the Ryukyu Islands probably consists of alien populations from mainland Asia and/ or mainland Japan. In the present study, we thus investigate the phylogenetic relationships of the East Asian C. c. chinesis/laeta subspecies group using mitochondrial and multi-locus nuclear DNA by Sanger and next-generation sequencing. We compare the phylogenetic relationships of mtDNA and nDNA, and clarify whether they are incongruent. We also conduct estimations of divergence time using mtDNA and nDNA. In addition, by using quantitative shell morphological analysis of these species groups, we test which of the phylogenies of mtDNA and nDNA refect the shell characteristics. Finally, we discuss the observed evolutionary patterns and provide information about these taxa that is of fundamental importance for their taxonomic classifcation. Results mtDNA phylogeny. Te results of maximum likelihood (ML) and Bayesian analyses were largely consistent with the relatively well-supported clades. Only clades with high support (i.e. posterior probability ≥0.95 or bootstrap support >70%) are here considered further. Cipangopaludina chinensis species were separated into four major clades (Fig. 1). Clade mtA was mainly composed of C. c. laeta from mainland Japan, though this clade was highly support by ML analysis only. However, we treat this group as monophyletic for convenience. It also included C. c. laeta of the Ryukyu Islands, except for Iriomote Island and Fukue Island. Te subspecies C. c. chinensis of Taiwan and China (Hunan), and two diferent species [C. longispira (Cl) and C. ventricosa] formed the clade mtB. In clade mtC, some populations of C. c. laeta from the northern part of Kyushu and C. c. chinensis of South Korea (Suncheon) were recovered as monophyletic. We recovered sister relationship between clades mtA–C and lineages of south Chinese species [including endemic species of ancient lake groups of Yunnan Province (Margarya) and C. longispira (KIZD148)]. However, some other populations of Kyushu, Vietnam, South Korea (Haenam), and Hawaii (introduced) clustered together as clade mtD; in addition, in comparison with the relationship between clades mtA–C and the south Chinese species group, clades mtA–C and clade mtD were more genetically diferent. As a consequence, the phylogenetic relationships refected the geographical distribution patterns (Fig. 2). Besides, clade mtD was divided into four subclades refecting the geographical distribution, but the Japanese populations (Kyushu Island) of this clade had only one haplotype (mtD4; Table S1). In contrast, clade mtA was distributed over a wide area but did not have any subclade refecting the geographical distribution pattern. Notably, a single haplotype (mtA3) was distributed from Hokkaido to Fukue Island.

Estimation of divergence time using mtDNA. For the molecular clock analysis, the ESS values produced in Tracer v.1.6 were considerably higher than 200. Each marginal likelihood estimate is UL model (−3299.3077) and SC model (−3313.7642), and the Bayes factor is 14.4566. For convenience, we frst assigned numbers to the cardinal nodes from 1 to 10 (Table 1, Fig. 3). Te topology of the tree almost completely coin- cided with that obtained for the Bayesian tree using MrBayes and the ML analyses (Fig. 1). Positive Bayes factor indicated UL model ft the data best. Te mean time of the frst divergence of the modern C. chinensis group (node 1) was estimated to be 5.53 Ma (UL model, lower and upper 95% HPD interval 3.10–8.41 Ma) and 5.33 Ma (SC model, lower and upper 95% HPD interval 3.89–6.90 Ma). In addition, the mean times for Clade mtA– mtC + South China (including Margarya) group (node 2) and Clade mtA + mtB + mtC (node 3) were estimated to be 3.63 Ma (UL model, lower and upper 95% HPD interval 2.00–5.73 Ma) and 2.88 Ma (SC model, lower and upper 95% HPD interval 2.05–3.72 Ma), and 2.37 Ma (UL model, lower and upper 95% HPD interval 1.18– 3.93 Ma) and 1.74 Ma (SC model, lower and upper 95% HPD interval 1.16–2.33 Ma), respectively. Tus, the last common ancestor of each clade diversifed in the Pliocene and Pleistocene eras. Additional information is shown in Table 1.

Genetic diferences of each population using nDNA. Estimated likelihood [LnP (D)] was found to be greatest when K = 2, suggesting that the four mitochondrial clades can be divided into two major clusters of SNPs in the frst STRUCTURE (Fig. 4a). Tese two clusters refect clades mtA and mtB–D. However, the lineage of MIG-seq SNP was not divided corresponding to the mitochondrial clades mtB–D, even when comparing hierarchically from K = 2 to K = 5. In addition, at least in four populations from the islands adjacent to Kyushu Island (Tsushima Island: Loc. No. 27, Tanegashima Island: 33, and Fukue Island: 35) and Iriomote Island (39), there are mixed genetic structures between the two major clusters. A scatter plot of the frst PCA also showed that these specimens can be divided into two major clusters refecting the result of the frst STRUCTURE (Fig. 4b). Te four populations, which have the mixed genetic structure as revealed by the frst STRUCTURE, showed positions relatively intermediate between those of the two major clusters in the plot. Tese results indicated that the two major genetic clusters could be regarded as C. c. laeta and C. c. chinensis. Terefore, for convenience, we treated these two clusters of MIG-seq SNP as laeta and chinensis in the second analyses. In the second STRUCTURE

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 2 www.nature.com/scientificreports/ www.nature.com/scientificreports

5 7/- mtA1 (37) mtA2* (8, 9, 38) mtA3* (1-4, 6, 7, 12, 15-17, 19, 21, 22, 35) mtA4 (5) mtA5 (14) mtA6* (36, 38) 7 3/- mtA7 (36) mtA mtA8 (18) - /94 mtA9 (23) 5 4/- mtA10 (23) mtA11* (23) 5 3/.66 mtA12 (11) mainland Japan mtA13 (37) Fukue Island mtA14* (10, 20) Ryukyu Islands

6 7/.93 mtB1* (42) mtB2 Cipangopaludina ventricoca Cv 8 1/.99 mtB3 (42) mtB mtB4* (43) 9 0/.99 C. longispira Cl Taiwan, China mtC1 (40) 7 4/.97 6 0/- mtC2 (28) 6 8/.99 mtC mtC3* (25, 28, 29, 34) 9 1/.99 South Korea mtC4 (26) Northern Kyushu Island C. longispira KIZD148 7 3/.99 Margarya dianchiensis KIZD43 M. monodi KIZD87 South China (Yunnan) 9 7/1.0 5 4/.89 9 5/.65 M. bicostata KIZD172 9 4/1.0 M. melanioides KIZD2 mtD1 (41) 8 3/.99 China, South Korea, Vietnam 9 9/1.0 mtD2* (44) Kyushu Island and its around areas 7 9/.99 mtD3* (44) mtD 100/1.0 100/1.0 mtD4*(24, 27, 30-33, 39) / C. chinensis chinensis KR31 mtD5 C. c. chinensis HI91 Heterogen longispira V148 100/1.0 C. japonica V254

100/1.0 S. quadrata quadrata LY01 S. purificata YC01 6 5/.99 5 4/.99 100/1.0 S. aeruginosa YT02 Outgroups -/.99 S. q. histrica V91 Anularya mansuyi KIZD21 100/1.0 Tchangmargarya yangtsunghaiensis T. multilabiata LJZ2014 0.04

Figure 1. Maximum likelihood consensus tree of East Asian viviparids based on combined sequences from the 16S and COI genes. Each operational taxonomic unit (OTU) label represents the haplotype number in Table S1. Samples from GenBank are indicated each species/subspecies name followed by the specimen ID. Te numbers in brackets next to the specimen ID indicate the sampling site number in Fig. 2. Each clade was diferentiated by a grey colour. Numbers on branches indicate maximum likelihood bootstrap values followed by Bayesian posterior probabilities. Scale bar indicates 0.04 substitutions per site. Te asterisk next to the OTU label indicates a haplotype shared by some other individuals.

analysis for C. c. laeta and C. c. chinensis, each estimated likelihood [LnP (D)] was also found to be greatest when K = 2 (Fig. 4a). Two SNP clusters of C. c. laeta refect the geographical distribution (one from the Ryukyu Islands, the other from mainland Japan except for Kyushu Island). Te populations of the Ryukyu Islands in particular were clearly diferent from other clusters, even when comparing hierarchically from K = 2 to K = 5. In addition, for the two SNP clusters of C. c. chinensis, one includes Japan (Kyushu Island and its around areas), while the second cluster includes other countries, refecting the geographical distribution. A scatter plot of the second PCA showed that each species is divided into two major groups, in line with the results of the second STRUCTURE (Fig. 4b). Moreover, two populations (Shiga: 15 and Kochi: 22) of C. c. laeta and three populations (Suncheon, South Korea: 40, Anhui, China: 41, and Taiwan: 43) of C. c. chinensis have a mixed genetic structure in the second STRUCTURE, and were shown to be located in positions relatively intermediate between those of the two major groups in the second PCA plots.

Results of population demography analysis. Te results of the population demography analysis by DIYABC using 160 SNPs showed that scenario 7 (Fig. 5) had the highest posterior probability (approximately 0.4) (Fig. S1). Estimated values for each posterior probability of scenarios are listed in Table S2. Based on this scenario, Pop1 and Pop3 initially separated from an unknown ancestral population, and subsequently Pop2 has been composed by Pop1 and Pop3. Te estimated parameters of scenario 7 are shown in Table S3. In the case of scenario 7, the median values of divergence times were 2.03 × 105 generations ago for time t1 (95% CI: 4.6 × 104–2.9 × 105) and 5.72 × 106 generations ago for time t2 (95% CI: 1.31 × 106–9.72 × 106) (Fig. S1). Some previous studies suggested that the sexual maturation of ‘C. chinensis’ is likely to take one year, with iteroparous reproduction for several years thereafer22–24. Terefore C. c. chinensis and C. c. laeta were separated from the unknown ancestral group before 5,720,000 years ago (t2), and the introgressive hybridisation [the median values of ra: 5.17 × 10−1 (95% CI: 3.94 × 10−1–6.41 × 10−1)] between the two species on Tanegashima Island occurred before 203,000 years ago (t1). Te result of PCA from Scenario 7 on model checking option is shown in Fig. S3. In addition, two

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 3 www.nature.com/scientificreports/ www.nature.com/scientificreports

1000 km

41 1 2

42 40

44 Japan (mainland) 3 Japan 43 (Ryukyu Islands) N 4 8 5, 6 Japan (mainland) 9 10 7 13 14 11 25 15 Tsushima 24 16 Island 23 26 12 17 27 36 28 18 35 Kyushu 19 Island 21 20 38 37 29 22 34 30 39 31 Fukue 32 Island 100 km Japan (Ryukyu Islands) 33 Iriomote Tanegashima Island Island

Figure 2. Map showing the sampling sites. Site numbers 1–39 are in Japan including the Ryukyu Islands; site 40 is in South Korea; 41 and 42 are in China; 43 is in Taiwan; and 44 is in Vietnam. See Table S1 for details. Te map was created using the sofware Adobe Illustrator CS6, Macintosh version, (https://www.adobe.com/jp/) and “Map data: Google, DigitalGlobe”.

Divergence Time using UL (Ma) Divergence Time using SC (Ma) Node No. Mean (Lower CI, Upper CI) Mean (Lower CI, Upper CI) 1 5.53 (3.10, 8.41) 5.33 (3.89, 6.90) 2 3.63 (2.00, 5.73) 2.88 (2.05, 3.72) 3 2.37 (1.18, 3.93) 1.74 (1.16, 2.33) 4 2.08 (0.75, 3.69) 1.59 (0.93, 2.29) 5 1.07 (0.35, 1.99) 0.77 (0.38, 1.21) 6 1.17 (0.28, 2.49) 0.64 (0.30, 1.02) 7 0.60 (0.08, 1.39) 0.33 (0.09, 0.61) 8 0.65 (0.04, 1.57) 0.46 (0.05, 1.04) 9 0.72 (0.09, 1.61) 0.57 (0.19, 0.99) 10 1.03 (0.36, 1.96) 0.67 (0.31, 0.96)

Table 1. Detailed results of divergence time estimation and geographic analyses.

diferent ‘hybrid’ scenarios (Scenarios 1 and 2) were chosen as the second and third scenarios (Fig. S1). Te type I and type II errors for scenario 7 was 0.554 and 0.436, respectively (Table S4).

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 4 www.nature.com/scientificreports/ www.nature.com/scientificreports

mtA3* mtA6* mtA8 mtA4 mtA12 mtA5 mtA7 mtA11* mtA mtA10 .87/.84 mtA9

.88/.92 mtA14* 10 mtA2* .62/.65 mtA1

.69/.75 mtA13 mtB1* .69/.72 mtB2 (Cipangopaludina ventricoca Cv) .99/.99 mtB4* mtB .99/.99 5 3 mtB3 mtB5 (C. longispira Cl)

.77/.80 mtC1 mtC2 .99/.99 .99/1.0 mtC 2 7 mtC3* mtC4 .98/.99 C. longispira KIZD148 8 Margarya dianchiensis KIZD43 .98/.99 .70/.78 .97/1.0 4 M. monodi KIZD87 1 .99/1.0 M. bicostata KIZD172 9 M. melanioides KIZD2 .99/.99 mtD2* .69/- mtD3* 1.0/1.0 mtD1 6 mtD .60/.93 mtD4 (C. c. chinensis KR31)* mtD5 (C. c. chinensis HI91) Miocene Pliocene Pleistocene 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 Ma

Figure 3. Maximum clade credibility tree generated with the BEAST2 analysis from the combined sequences from the 16S and COI genes. Te outgroups (Anularya, Cipangopaludina japonica, Heterogen, Sinotaia, and Tchangmargarya) are not shown. Each operational taxonomic unit (OTU) label represents the haplotype number in Table S1. Samples from GenBank are indicated each species/subspecies name followed by the specimen ID. Each clade was diferentiated by a grey colour. Numbers on branches indicate Bayesian posterior probabilities of UL model followed by Bayesian posterior probabilities of SC model. Node bars indicate a 95% CI for the divergence times by UL model analysis. Te principal nodes are named with nominal numbers. Te asterisk next to the OTU label indicates a haplotype shared by some other individuals.

Morphological variation. Based on the results of PCA using the Fourier coefcients of all 157 specimens, 11 efective PCs were chosen by SHAPE v1.3 (Table S5). Te frst CDA was performed based on the 11 PCs and on shell width (Wilks’ lambda = 0.280, F = 6.287, P < 0.001; Fig. 6a; Table S6). Te frst and second canonical components explained 77.58% and 90.21% of the total variance, respectively. Te second CDA was performed based on the same 11 PCs and on shell width (Wilks’ lambda; value = 0.412, F = 17.15, P < 0.001; Fig. 6b; Table S6). Te frst canonical components explained 100% of the total variance. Te frst and second CDA results showed that there was little overlap in shell morphology between C. c. chinensis and C. c. laeta; nevertheless, clades mtB–D largely overlapped.

Discussion. Enigmatic incongruence between mtDNA and nDNA was found in this study. Two closely related mitochondrial clades (mtB and mtC) and a diferent clade (mtD) clustered into a monophyletic group in the phylogenetic relationships obtained from the nDNA (Fig. 4). In addition, the shell morphological pattern refected the phylogenetic relationship of nDNA more than that of the mtDNA (Fig. 6). Terefore, the specimens used in this study can be separated into two diferent subspecies: C. c. chinensis and C. c. laeta. In general, for a number of reasons, gene trees sometimes depart from species trees. Explanations for this include the presence of cryptic species25,26, incomplete lineage sorting27, and introgressive hybridisation28. Dealing with the incongruence between genes is a central issue in phylogeny29, but the cause of the incongruence in this study is still unclear. However, our results indicate that the phylogenetic incongruence might have occurred through introgressive hybridisation. Te results of divergence time estimation using mtDNA (node 1) are almost included in the times- cale of t2 of the ABC analysis, suggesting that these two species diversifed in the Pliocene to early Pleistocene era (Fig. 3; Table S3). In addition, the results of DIYABC also indicate that these two subspecies C. c. chinensis and C. c. laeta actually undertook introgressive hybridisation (t1) afer the divergence of clades mtA–C (node 3). According to a previous study, genetic information of extinct species can be included in those of modern species30. Terefore, hybridisation with extinct species (or population) occurred, which might have caused the incongruence. As an alternative hypothesis, considering possibility of hybridisation between two extant species, ancient introgressive hybridisation between the last common ancestors of clades mtA–C and mtD may also be

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 5 www.nature.com/scientificreports/ www.nature.com/scientificreports

First analysisSecond analysis a laeta chinensis

mt Ryukyu Other mtA mtCmtD mainland Japan Japan B Islands Countries 2 1 9 3 9 2 3 1 26 38 37 36 35 23 22 15 13 11 28 44 41 42 43 29 32 40 34 33 30 31 27 24 25 39 38 37 36 23 22 15 13 11 40 42 41 43 32 30 24 44 31 25 28 29 34 26 K=2

K=3

K=4

K=5

b First analysisSecond analysis 6 5 laeta 5 chinensis 4 2 ) 0 0 0 -2 -4 PC2 (4.78% PC2 (11.49%) -5 PC2 (11.03%) -6 -5 -8 mixed genetic individual mixed genetic individual mixed genetic individual -10 -15-10 -5 0510 -10-50 5 -15-10 -5 05 PC1 (48.93%) PC1 (24.89%) PC1 (15.01%)

Figure 4. Results of the population genetic analyses. (a) Results of the STRUCTURE analyses for K = 2–5. Te grey bars and numbers above the results of the STRUCTURE indicate each mitochondrial clade and distribution area, and sampling site number in Fig. 2, respectively. Te green and blue colours indicate the two diferent genetic clusters. (b) Plots of results from the principal component analyses (PCA) based on MIG-seq SNPs. Te green and blue, and grey symbols under the plots indicate the two genetic clusters and the mixed genetic individuals of STRUCTURE for K = 2, respectively.

the cause of the incongruence. Afer that ancient introgressive hybridisation, the last common ancestors of clades mtA–C would have been divided into three diferent clusters by geographical events. In any case, the mechanism by which specifc genotypes spread is unclear. For example, it remains unresolved why the genetic information of the extinct species cannot be found in the nDNA. A possible explanation of this is diferences in population size. In addition, non-neutral mtDNA may afect the observed incongruence. mtDNA is related to respiration and temperature adaptations31–33, so relatively adaptive mtDNA types for each habitat and environment may remain and spread34. In the case of freshwater fsh, one mtDNA haplotype spread afer introgressive hybridisation35. Some phylogenetic studies using freshwater molluscs showed a complex pattern of

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 6 www.nature.com/scientificreports/ www.nature.com/scientificreports

Scenario 1 Scenario 2 Scenario 3

t1 ra 1-ra 1-ra ra

0 Pop1 Pop2 Pop3 Pop3 Pop2 Pop1 Pop3 Pop2 Pop1 N1 N2 N3 Scenario 4 Scenario 5 Scenario 6

0 Pop2 Pop3 Pop1 Pop2 Pop1 Pop3 Pop1 Pop2 Pop3

Scenario 7 Scenario 8 Scenario 9

N1 N2 t1 N3 ra 1-ra NA

0 Pop1 Pop2 Pop3 Pop3 Pop2 Pop1 Pop2 Pop1 Pop3

Figure 5. Simulated nine scenarios. Pop1 is Sasebo (V1263–1265), Pop2 is Tanegashima Island (V632, V1416, V1417), and Pop3 is Sapporo (V983, V984). N1, N2 N3 and NA is population size. t1 and t2 is generation time for merging of population.

mtDNA phylogeny13,21,36. Taking the above reason into account, the phylogeny of some freshwater organisms is probably associated with whether the mtDNA haplotype is adaptive to each habitat and environment. Moreover, directionality in producing a juvenile may afect the patterns of mtDNA variations. For example, in the case of crossbreeding between female C. c. laeta and male C. c. chinensis, C. c. laeta probably has its juveniles, while in the opposite pattern, C. c. chinensis cannot have its juveniles. If directional crossbreeding between the extinct and modern species occurred, it may have resulted in the incongruence shown in this study. In any case, to compre- hensively address the question whether introgressive hybridization occurred, and to clarify the role of hybridization in genetic divergence and the mechanism of introgression, additional population genetic analysis using more individuals of these two species and a crossbreeding experiment are needed. Whatever the cause of the incongruence, the high dispersal ability of these snails may also be related to their unique phylogeny. Notably, each haplotype of mtA, C, and B is distributed across a relatively wide area (Figs 1 and 2; Table S1). Te cause of such rapid expansion in their distribution range is unclear, but the geographical distance and phylogenetic relationship of both species do not necessarily coincide, even when considering the trees of nDNA; therefore, some populations may be formed by human activity. In fact, C. c. chinensis and/or C. c. laeta have been introduced into North America by human activity over the last two centuries37, and also into Europe38, as well as alien populations becoming established within Japan: a population of Iriomote Island might have been introduced from population of western part of mainland Japan (Yamaguchi Prefecture)39. In addition, this genus including C. c. chinensis and C. c. laeta inhabits various natural and artifcial freshwater environments, including shallow lakes, streams, wetlands, and ponds, as well as rice and taro farms40–42. Originally, these species inhabited the natural habitat, but in general, these habitats are unstable and the acidity readily increases due to the accumulation of plant remains. Shells are easily decomposed in highly acidic water43, so few fossils of these species might have been obtained in Japan.

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 7 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 6. Results of canonical discriminant analysis (CDA). Green and blue indicate laeta and chinensis, respectively. (a) Plots of results from the frst CDA based on shell width – D and PCs. Te symbols under the plots indicate the taxa of each mitochondrial clade. (b) Histogram of results of the second CDA. Te photographs on the lef indicate each shell morphology. N indicates the number of individuals.

While these species may have high dispersal ability, C. c. laeta of the Ryukyu Islands is genetically diferent from that of mainland Japan, and is grouped together in an island-specifc manner. In clade mtA, only one individual of the Ryukyu Islands (V656) shares a haplotype (mtA2) with that of mainland Japan (Yamagata), but the other individuals do not have a haplotype that is broadly distributed in mainland Japan (Fig. 1). Terefore, our results suggest that the native and endemic populations of C. c. laeta may be distributed in the Ryukyu Islands. However, the reason why C. c. laeta was not found in mainland Kyushu is unclear. Te results of nDNA showed that the populations on the islands adjacent to Kyushu Island (Tanegashima) were composed of hybrids between C. c. laeta and C. c. chinensis (Fig. 5, scenario 7); therefore, C. c. laeta might have been distributed in the mainland of Kyushu. If the populations of C. c. laeta on the Ryukyu Islands are native, C. c. laeta was once widely distributed in Japan, including on Kyushu Island, but possibly became extinct due to some large-scale disturbance such as a volcanic eruption. In fact, repeated volcanic eruptions are known to have occurred on Kyushu Island or its adjacent islands44–46. For example, a bottleneck event followed by population expansion was suggested to have occurred for the Japanese macaque (Macaca fuscata yakui) on Yakushima Island47. Alternatively, C. c. chinensis may have expanded its distribution on the Kyushu mainland and eradicated C. c. laeta by interspecifc competi- tion. Additional feld surveys are needed to obtain a deeper understanding of the distribution patterns of these two species. Te present fndings also raised issues about the taxonomy of these species. Before the work of48, the taxon ‘laeta’ as used in the present study was treated as Paludina malleata, but48 indicated that the following three species are synonyms of laeta: Paludina malleata (Reeve, 1863) and P. abbreviata (Reeve, 1863), and Viviparus stelmaphora (Bourguignat, 1862), described from Japan and China, respectively. In addition, C. c. laeta was also determined to be a subspecies of C. c. chinensis in the work of48. However, the detailed type localities of C. c. laeta, P. malleata , and P. abbreviata were not identifed in the descriptions. According to49, C. c. laeta was described from a specimen or specimens collected by Siebold, who lived in Nagasaki (on the mainland of Kyushu), so the type specimens of C. c. laeta are probably C. c. chinensis. However, it is difcult to determine whether this ‘laeta’ is C. c. chinensis as used in this study using only information on the shell. Nevertheless it may be reasonable to treat ‘laeta’ of Japan except for Kyushu Island as C. malleata, afer this study, though the type locality of P. mal l eata is unclear.

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 8 www.nature.com/scientificreports/ www.nature.com/scientificreports

In present study, we treated as chinensis for continental species, but actually some diferent species were described in the continental50. Diferent taxa are included in the continental Cipangopaludina (e.g. C. longispira, C. ventricosa) in GenBank. However, taxon name used in the GenBank data may include misidentifcation, because the two C. longispira belong to diferent clades respectively (Fig. 1). Alternatively, they probably include cryptic species. Similarly, Cipangopaludina ventricosa from GenBank can be also treated as misidentifcation, or it can be a junior synonym of C. chinensis if it is not misidentifcation. Moreover, these taxonomic issues are probably associated with phenotypic plasticity of shell morphology21. In any case, the taxonomic issue surroundings C. chinensis may involve another continental species. Some phylogenetic studies using continental species were conducted13,51,52. However comprehensive phylogenetic relationship among continental species is still unclear. So we need to perform more sampling from China and some countries of mainland East Asia, and phylogenetic analysis using such samples. In particular, further study using SNPs including Margarya and its closely related taxa is needed for clarifying the paraphyly of C. chinensis. Our study revealed a complex evolutionary pattern, including phylogenetic incongruence. We clearly identifed the occurrence of introgressive hybridisation of Viviparidae for the frst time. In addition, we focused on the role of hybridisation in genetic diversifcation and adaptive radiation53,54. In fact, Margarya is closely related to laeta and chinensis20 (Fig. 1), which implies the possibility that the genetic divergence of this genus occurred via hybridisation13. Combined analyses using both mtDNA and nDNA, and both simple marker and multi-locus approaches, are useful for clarifying the evolutionary history of these viviparid snails. Materials and Methods Samples. Sampling was performed in the Japanese Archipelago including the Ryukyu Islands, China, Taiwan, and Vietnam. In total, we collected 210 individuals (Fig. 2, Table S1). Since48, Japanese subspecies of C. chinensis has been treated as a subspecies C. c. laeta. Terefore, for convenience, C. chinensis collected from Japan was treated as subspecies laeta, whereas snails of other countries were nominotypical subspecies chinensis in the present study. A fragment of the foot muscle from each individual was stored in 99.5% ethanol for DNA extraction. We also obtained GenBank data of C. c. chinensis (HI91 and KR31). Moreover, we used GenBank data of Cipangopaludina longispira, Cipangopaludina ventricosa, Margarya bicostata, Margarya dianchiensis, Margarya melanoides, and Margarya monodi as ingroup for mitochondrial phylogenetic analysis because these species are morphologically and/or genetically closely related to the C. chinensis group13,21,50. Four Asian genera of Viviparidae, Anularya mansuyi, Cipangopaludina japonica, Heterogen longispira, Tchangmargarya multilabiata, T. yangtsunghaiensis, Sinotaia aeruginosa, Sinotaia purifcata, Sinotaia quadrata quadrata, and Sinotaia quadrata histrica, were also obtained from GenBank as outgroups21.

Mitochondrial phylogenetic analysis. Total DNA was extracted in accordance with the work of 21. Fragments of the mitochondrial cytochrome oxidase subunit I (COI) and 16S rRNA genes were amplified using the primers LCO1490 (5′-GGTCAACAATCATAAAGATATTGG-3′) and HCO2198 (5′-TAAACTTCAGGGTGACCAAAAAATC-3′)55 for COI and 16S-arL (5′-CGCCTGTTTAACAAAAACAT-3′) and 16S-brH (5′-CCGGTCTGAACTCAGATCACGT-3′)56 for 16S. PCR reactions were conducted under the conditions described by21. PCR products were purified using Exo-SAP-IT (Amersham Biosciences, Little Chalfont, UK). Sequencing was performed using the PCR primers and BigDyeTM Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, CA, USA), and electrophoresis was carried out using an ABI 3130xl sequencer (Applied Biosystems). Te newly generated sequences have been deposited in the GenBank databases (Table S1). Alignment of the COI sequences was straightforward and required no gaps; 16S sequences were aligned using MUSCLE57. GBLOCKS v0.91b58 was used to select regions in the aligned sequences that were confdently aligned for analysis (Table S7). Afer sequence selection, we summarised the same sequence alignments because of node density artefacts59,60. Phylogenetic analyses were conducted for the combined data set using maximum likelihood (ML) and Bayesian Inference. For ML and Bayesian analyses, we used Kakusan4-4.0.2011.05.2861 to select the appropriate models for sequence evolution (Table S7). Using the selected models, we performed ML analysis in RAxML HPC262. Nodal support for ML analysis was assessed using bootstrap analyses with 1000 replications. Bayesian analysis was performed in MrBayes v3.1.263 using two simultaneous runs, consisting of four simultaneous chains for 30 million generations and sampling the trees every 1000 generations. Te frst 20% of trees of each run were discarded as burn-in. Both ML and Bayesian analyses were performed at the San Diego Supercomputer Center through the CIPRES Science Gateway64.

Estimation of divergence time. Approximate divergence times were estimated using the two diferent clock models (strict clock [SC] and uncorrelated log-normal relaxed clock [UL]) as implemented in BEAST 2.4.565 at the San Diego Supercomputer Center through the CIPRES Science Gateway64. We used the Bayes factor based on marginal likelihood estimates to compare each clock model66. We used the default parameters except for step number (we set 100 steps). Marginal likelihood estimates were obtained using Path Sampler (included in the BEAST package). Te model of COI was chosen from models in which the molecular clock rate based on some fossil records of viviparid snails is considered in67. A molecular clock rate (uniform prior) ranging from 0.0068 to 0.0118 (substitutions per site and My) was proposed for the COI gene of the family Viviparidae in67. We applied the HKY + G model to both COI and 16S sequences, respectively. In addition, we set the parameters of the site models to Gamma category count = 4 and shape = 1.0 as described by68. Te Yule process was used to model speciation. Te Monte Carlo Markov chain was run four times for 30 million generations with sampling every 1000 generations to ensure that the efective sample size (ESS) values were above 1000 for all parameters. In total, we performed 4 runs (both UL and SC). Afer that, the frst 20% of trees of each run were discarded as

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 9 www.nature.com/scientificreports/ www.nature.com/scientificreports

burn-in, and the remaining trees were combined to produce an ultrametric consensus tree using Logcombiner and Treeannotator v1.5.3 (included in the BEAST sofware package).

MIG-seq analysis and SNP detection. Molecular markers of nDNA generally used for phylogenetic analysis, such as 18S rRNA, 28S rRNA, and histone 3 (H3) genes, have slower divergence rates than mtDNA, and are not suitable for clarifying details of the genetic relationship between C. c. chinensis and C. c. laeta (Hirano et al., in prep). Terefore, we obtained genome-wide SNP information of these species groups by multiplexed inter-simple sequence repeat (ISSR) genotyping by sequencing (MIG-seq69) from total DNA. MIG-seq is a microsatellite-associated DNA sequencing technique—a type of reduced representation sequencing that includes restriction site-associated DNA sequencing (RAD-seq)70. Tis method is especially efective for low-quality DNA and small quantities of DNA71. Preparation of the MIG-seq library was performed under a method modifed from69 on an Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA) (Tsunamoto et al., unpublished), using an MiSeq Reagent Kit v3 (150 cycle; Illumina). Removal of the primer regions and quality fltering were conducted in accordance with the work of 69. SNPs were called using the Stacks sofware package version 1.4172 for further population genetic analyses. We used Stacks with the following parameters: maximum distance between stacks (M) = 1, maximum distance allowed to align secondary reads to primary stacks (N) = 1, and minimum depth option = 10 (-m 10). In addition, we also used a gapped option. We selected only SNPs recorded at a rate of more than 50% among samples for MIG-seq. We also excluded individuals with a high defciency rate (≥50%) from the analysis, with the exception of an individual from South Korea (M22436). As a results, 1153 SNPs among all populations were detected.

MIG-seq population genetic analyses. To compare the results of mtDNA with the genetic structure of nDNA, we estimated individual genotypes of nDNA with STRUCTURE v2.3.473,74, based on the MIG-seq data set (Table S1). Te number of preassigned genetic clusters (K) was assumed to range from 1 to 5. We performed 10 independent runs for each K value. Each run included 10,000 burn-in iterations and 10,000 iterations. To help determining the optimal K, ∆K was calculated as described by74 using Structure harvester web v0.6.9475. Bar charts for the proportions of the membership coefcient of each individual in STRUCTURE analysis in 10 runs for each K were summarised using CLUMPP v1.1.276 and visualised in distruct v1.177. In addition, to determine the genetic structure of nDNA, we also conducted a principal component analysis (PCA) with GenoDive v 2.0b2778 as well as STRUCTURE analysis. When we performed the frst STRUCTURE and PCA, we used 75 specimens (1153 SNPs). Next, we separately conducted the second STRUCTURE and PCA analyses using 33 specimens (997 SNPs) of chinensis and 32 specimens (685 SNPs) of laeta, respectively, excluding the mixed genetic population between the two species as estimated by the frst STRUCTURE described above. Tis is because we investigate genetic structure within each genetic group determined by the frst STRUCTURE and PCA. To confrm the history of genetic diversifcation, we used an Approximate Bayesian Computation (ABC) approach in DIYABC v.2.179. ABC provides estimates of demographic and historical parameters and quantitative comparisons of evolutionary scenario80,81. On the basis of the structure analysis and PCA analysis on the SNPs, populations were grouped into three groups: Pop1 [chinensis: Sasebo (28); V1263–1265], Pop2 [mixed: Tanegashima Island (33); V632, V1416, V1417], and Pop3 [laeta: Sapporo (2); V983, V984]. Because the results of STRUCTURE showed that some populations appear to have a mixed genetic structure between two diferent genetic groups, the above populations (Pop1–3) were used to investigate how the genetic structure was created. Considering isolation among the groups and an unknown ancestral group (NA), nine scenarios were assumed, as shown in Fig. 5. We conducted the pilot run with wide extensive parameters and all summary statistics values. We used adopted the following parameters based on the results from the pilot run: maximum efective numbers of population size N1 (population size of Pop1) and N3 (population size of Pop3), and of N2 (population size of Pop2) were set to 500,000 and 100,000, respectively. In addition, NA (population size of PopNA) was set to 5,000,000. Maximum efective number of generations was set to 300,000 (t1: time when the newer event occurred) and 10,000,000 (t2: time when the older event occurred). Te parameter settings are listed in Tables S8. In total, we used 6 summary statistics (each Mean of non-zero values and Variance of non-zero values of Genetic diversities, Fst distances, and Nei’s distances). We compared the diferent scenarios by calculating their relative posterior probabilities using a logistic regression method from the 1% of simulated data sets most closely resembling the observed data set. We used the option ‘confdence in scenario choice’ in DIYABC to evaluate the validity in scenario choice. We calculated each scenario specifc prior based error using simulated 500 datasets per scenario with the same parameter setting. Under the most likely scenario as determined by the logit transformation of parameters, the posterior distributions of parameters were estimated on the 1% of simulated data sets most closely resembling the observed data set. We also used the option ‘model checking’ with principal component analysis (PCA) using DIYABC to assess the goodness-of-ft of the three scenarios. Tis option can be used to evaluate the consistency of the observed data with the posterior predictive distribution of the model for the best scenario.

Morphological analysis. To determine whether the phylogenetic relationships of mtDNA and nDNA among the chinensis/laeta subspecies correspond with the diferences in shell morphology, we analysed the variation in shell shape, along the entire outline of the shell, except for the inside aperture and suture, as well as shell width (D; Fig. 7). We used adult shells that have strongly thickened outer lips. In total, we undertook the analyses on 157 specimens (80 individuals of C. c. chinensis and 77 individuals of C. c. laeta), excluding the mixed genetic population between the two species as estimated by the frst STRUCTURE described above. First, we took a photograph of the shell of each specimen obtained from the feld and measured the shell width using digital caliper (Tables S1 and S5). For the quantitative evaluation of entire shell shape, we generated elliptic Fourier descriptors (EFDs82) using the same digital images taken by us. All of the images used for the

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 10 www.nature.com/scientificreports/ www.nature.com/scientificreports

Figure 7. Examples of fgure data for the morphological analysis. (a) Shell morphology and shell width (D). (b) Te entire outline of the shell.

morphological analysis were placed so as to face the shell aperture forward. Te parameter of harmonic ampli- tudes was set to n = 40. For the analysis of shell shape using Fourier coefcients obtained from the EFDs, we conducted a principal component analysis (PCA). We used SHAPE v1.3 to process the digital images, obtain EFDs, and perform PCA83. Second, we conducted a canonical discriminant analysis (CDA) based on PCs obtained from PCA using the EFDs and shell width. Te number of efective PCs was calculated as the number of PCs that account for a pro- portion of the variability larger than 1/number of PCs analysed using SHAPE ver. 1.3. Te CDAs were conducted with XLSTAT (Addinsof, Paris, France). When we performed the frst CDA, these scores were compared for each mtDNA clade. Next, to investigate the relationship between the lineage of nDNA (MIG-seq SNPs) and shell morphology, we conducted a second CDA. References 1. Sun, Z. et al. Rapid and recent diversifcation patterns in Anseriformes birds: Inferred from molecular phylogeny and diversifcation analyses. PLoS One 12, e0184529 (2017). 2. Walther, A. C., Lee, T., Burch, J. B. & Ó Foighil, D. Confrmation that the North American ancylid Ferrissia fragilis (Tryon, 1863) is a cryptic invader of European and East Asian freshwater ecosystems. Journal of Molluscan Studies 72, 318–321 (2006). 3. Haase, M. & Zielske, S. Five new cryptic freshwater gastropod species from New Caledonia (Caenogastropoda, Truncatelloidea, Tateidae). ZooKeys 523, 63–87 (2015). 4. Joly, S., McLenachan, P. A. & Lockhart, P. J. A statistical approach for distinguishing hybridization and incomplete lineage sorting. American Naturalist 174, 54–70 (2009). 5. Parham, J. F. et al. Genetic introgression and hybridization in Antillean freshwater turtles (Trachemys) revealed by coalescent analyses of mitochondrial and cloned nuclear markers. Molecular Phylogenetics and Evolution 67, 176–187 (2013). 6. Kubatko, L. S. Identifying hybridization events in the presence of coalescence via model selection. Systematic Biology 58, 478–488 (2009). 7. Yu, Y., Tan, C., Degnan, J. H. & Nakhleh, L. Coalescent histories on phylogenetic networks and detection of hybridization despite incomplete lineage sorting. Systematic Biology 60, 138–149 (2011). 8. Takahashi, T. & Moreno, E. A RAD-based phylogenetics for Orestias fishes from Lake Titicaca. Molecular Phylogenetics and Evolution 93, 307–317 (2015). 9. Razkin, O. et al. Species limits, interspecifc hybridization and phylogeny in the cryptic land snail complex Pyramidula: the power of RADseq data. Molecular Phylogenetics and Evolution 101, 267–278 (2016). 10. Rivers, D. M., Darwell, C. T. & Althof, D. M. Phylogenetic analysis of RAD-seq data: examining the infuence of gene genealogy confict on analysis of concatenated data. Cladistics 32, 672–681 (2016). 11. Rosenberg, G. A new critical estimate of named species-level diversity of the Recent Mollusca. American Malacological Bulletin 32, 308–322 (2014). 12. Chiba, S. Accelerated evolution of land snails Mandarina in the oceanic Bonin Islands: evidence from mitochondrial DNA sequences. Evolution 53, 460–471 (1999). 13. Du, L., Yang, J., Rintelen, T., von., Chen, X. & David, A. Molecular phylogenetic evidence that the Chinese viviparid genus Margarya (Gastropoda: Viviparidae) is polyphyletic. Chinese Science Bulletin 58, 2154–2162 (2013). 14. Miura, O., Urabe, M., Nishimura, T., Nakai, K. & Chiba, S. Recent lake expansion triggered the adaptive radiation of freshwater snails in the ancient Lake Biwa. Evolution Letters 3, 43–54 (2019). 15. Aksenova, O. V. et al. Species richness, molecular taxonomy and biogeography of the radicine pond snails (Gastropoda: Lymnaeidae) in the Old world. Scientifc Reports 8, 11199 (2018). 16. Saito, T. et al. Endangered freshwater limpets in Japan are actually alien invasive species. Conservation Genetics 19, 947–958 (2018). 17. Ministry of Environment. Red Data Book 2014 -Treatened Wildlife of Japan- Volume 6, Mollusks (GYOSEI Corporation, Tokyo, 2014). 18. Kurozumi, T. Natural environment and human life of Amami and Okinawa from the viewpoint of shells. In Kinoshita, N. (eds) Subsistence and exchange in prehistoric Ryukyus based on excavations in Amami and Okinawa. (Kinoshita lab., Faculty of letters, Kumamoto University, Kumamoto, 2003). 19. Kurozumi, T. Study on the present status and conservation of natural environment of Tama-gawa River based on the molluscs. (Te Tokyu Foundation for Better Environment, Tokyo, 2002). 20. Matsuoka, K. & Shimizu, I. Freshwater Molluscs from the Middle Pleistocene Toyono Formation in Nagano Prefecture, Central Japan. Bulletin of the Mizunami Fossil. Museum 14, 89–102 (1988).

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 11 www.nature.com/scientificreports/ www.nature.com/scientificreports

21. Hirano, T., Saito, T. & Chiba, S. Te phylogeny of freshwater viviparid snails in Japan. Journal of Molluscan Studies 81, 435–441 (2015). 22. Stanczykowska, A., Magnin, E. & Dumouchel, A. Etude de trois populations de Viviparus malleatus (Reeve) de la region de Montreal. I. Croissance, fecondite, biomasse et production annuelle. Canadian Journal of Zoology 49, 1431–1441 (1971). 23. Jokinen, E. Cipangopaludina chinensis (Gastropoda: Viviparidae) in North America. Nautilus 96, 89–95 (1982). 24. Terriault, T. & Kott, E. Cipangopaludina chinensis malleata (Gastropoda: Viviparidae) in southern Ontario: An update of the distribution and some aspects of life history. Malacological Review 35–36, 111–121 (2002–2003). 25. Minton, R. L. & Lydeard, C. Phylogeny, taxonomy, genetics and global heritage ranks of an imperiled, freshwater snail genus Lithasia (Pleuroceridae). Molecular Ecology 12, 75–87 (2003). 26. Kim, W. J. et al. Systematic relationships of Korean freshwater snails of Semisulcospira, Koreanomelania, and Koreoleptoxis (Cerithiodiea; Pleuroceridae) revealed by mitochondrial Cytochrome Oxidase I sequences. Korean Journal of Malacology 26, 275–283 (2010). 27. Glaubrecht, M. & Köhler, F. Radiating in a river: systematics, molecular genetics and morphological diferentiation of viviparous freshwater gastropods endemic to the Kaek River, central Tailand (Cerithioidea, Pachychilidae). Biological Journal of the Linnean Society 82, 275–311 (2004). 28. Glaubrecht, M. & von Rintelen, T. Te species focks of lacustrine gastropods: Tylomelania on Sulawesi as models in speciation and adaptive radiation. Hydrobiologia 615, 181–199 (2008). 29. Scornavacca, C. & Galtier, N. Incomplete lineage sorting in mammalian phylogenomics. Systematic Biology 66, 112–120 (2017). 30. Lee, T. et al. Tahitian tree snail mitochondrial clades survived recent mass extirpation. Current Biology 17, R502–R503 (2007). 31. Nagao, Y. et al. Decreased physical performance of congenic mice with mismatch between the nuclear and the mitochondrial genome. Genes & Genetic Systems 73, 21–27 (1998). 32. Itoi, S., Kinoshita, S., Kikuchi, K. & Watabe, S. Changes of carp F0F1-ATPase in association with temperature acclimation. American Journal of Physiology 284, R153–R163 (2003). 33. Mishmar, D. et al. Natural selection shaped regional mtDNA variation in humans. Proceedings of the National Academy of Sciences of the United States of America 100, 171–176 (2003). 34. Ballard, J. W. & Whitlock, M. C. Te incomplete history of mitochondria. Molecular Ecology 13, 729–744 (2004). 35. Mukai, T. & Takahashi, H. Phylogeography with hybridization: species and distribution pattern. In: Watanabe, K. & Takahashi, H. (Eds), Natural history of freshwater fsh geography (Hokkaido University Press, Sapporo, 2010). 36. Miura, O. et al. Rare, divergent Korean Semisulcospira spp. mitochondrial haplotypes have Japanese sister lineages. Journal of Molluscan Studies 79, 86–89 (2013). 37. Abbot, R. T. Snail invader. Natural History 59, 80–85 (1950). 38. Soes, D. M., Majoor, G. D. & Keulen, S. M. A. Bellamya chinensis (Gray, 1834) (Gastropoda: Viviparidae), a new alien snail species for the European fauna. Aquatic Invasions 6, 97–102 (1950). 39. Kitano, T., Watanabe, K., Sakihara, K., Hojyo, Y. & Kono, H. Utilization of Cipangopaludina chinensis laeta in Iriomote Island and the origin of the population found in the Minapishi area of the western part. Te study review of Iriomote Island 2014, ORRC, Tokai University, 44–49 (2015). 40. Pace, G. L. Te freshwater snails of Taiwan (Formosa). Malacological Review Supplement 1, 1–117 (1973). 41. Chen, L. C. Aquaculture in Taiwan (Fishing News Books, Oxford, 1990). 42. Chiu, Y., Chen, H., Lee, S. & Chen, C. A. Morphometric analysis of shell and operculum variations in the Viviparid snail, Cipangopaludina chinensis (Mollusca: Gastropoda), in Taiwan. Zoological Studies 41, 321–331 (2002). 43. Bednarsek, N. et al. Limacina helicina shell dissolution as an indicator of declining habitat suitability due to ocean acidifcation in the California Current Ecosystem. Proceedings of the Royal Society B: Biological Sciences 281, 20140123 (2014). 44. Okuno, M. Chronology of tephra layers in southern Kyushu, SW Japan, for the last 30,000 years. Te Quaternary research 41, 225–236 (2002). 45. Machida, H. & Arai, F. Akahoya Ash – A Holocene widespread tephra erupted from the Kikai Caldera, south Kyushu, Japan. Te Quaternary research 17, 143–163 (1978). 46. Machida, H. & Arai, F. Atlas of Tephra in and around Japan [revised edition]. (Te University of Tokyo Press, Tokyo, 2003). 47. Hayaishi, S. & Kawamoto, Y. Low genetic diversity and biased distribution of mitochondrial DNA haplotypes in the Japanese macaque (Macaca fuscata yakui) on Yakushima Island. Primates 47, 158–164 (2006). 48. Habe, T. Te type specimens of Cyclophorus herklotsi and Cipangopaludina chinensis laeta. Te Chiribotan 9, 46 (1976). 49. Pilsbry, H. A. Revision of Japanese Viviparidae, with Notes on Melania and Bithynia. Proceedings of the Academy of Natural Sciences of Philadelphia 54, 115–121 (1902). 50. Lu, H. F., Du, L. N., Li, Z. Q., Chen, X. Y. & Yang, J. X. Morphological analysis of the Chinese Cipangopaludina species (Gastropoda; Caenogastropoda; Viviparidae). Zoological. Research 35, 510–527 (2014). 51. Zhang, L. J., Chen, S. C., Yang, L. T., Jin, L. & Köhler, F. Systematic revision of the freshwater snail Margarya Nevill, 1877 (Mollusca: Viviparidae) endemic to the ancient lakes of Yunnan, China, with description of new taxa. Zoological Journal of the Linnean Society 174, 760–800 (2015). 52. Wang, J. G., Zhang, D., Jakovlić, I. & Wang, W. M. Sequencing of the complete mitochondrial genomes of eight freshwater snail species exposes pervasive paraphyly within the Viviparidae family (Caenogastropoda). PLoS ONE 12, e0181699 (2017). 53. Seehausen, O. Hybridization and adaptive radiation. Trends in ecology and evolution 19, 198–406 (2004). 54. Kagawa, K. & Takimoto, G. Hybridization can promote adaptive radiation by means of transgressive segregation. Ecology Letters 21, 264–274 (2018). 55. Folmer, O., Black, M., Hoeh, W., Lutz, R. & Vrijenhoek, R. DNA primers for amplifcation of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Molecular Marine Biology and Biotechnology 3, 294–299 (1994). 56. Palumbi, S. R. et al. Te Simple Fool’s Guide to PCR. (Department of Zoology and Kewalo Marine Laboratory, University of Hawaii, Honolulu, 1991). 57. Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Research 32, 1792–1797 (2004). 58. Castresana, J. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Molecular Biology and Evolution 17, 540–552 (2000). 59. Fitch, W. M. & Bruschi, M. Te evolution of prokaryotic ferredoxins—with a general method correcting for unobserved substitutions in less branched lineages. Molecular Biology and Evolution 4, 381–394 (1987). 60. Fitch, W. M. & Beintema, J. J. Correcting parsimonious trees for unseen nucleotide substitutions: Te efect of dense branching as exemplifed by ribonuclease. Molecular Biology and Evolution 7, 438–443 (1990). 61. Tanabe, A. S. K4 and Aminosan: two programs for comparing nonpartitioned, proportional and separate models for combined molecular phylogenetic analyses of multilocus sequence data. Molecular Ecology Resources 11, 914–921 (2011). 62. Stamatakis, A. RAxML-VI-HPC: maximum likelihood-based phylogenetic analyses with thousands of taxa and mixed models. Bioinformatics 22, 2688–2690 (2006). 63. Ronquist, F. & Huelsenbeck, J. P. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19, 1572–1574 (2003). 64. Miller, M., Pfeifer, W. & Schwartz, T. Creating the CIPRES Science Gateway for inference of large phylogenetic trees. In Gateway Computing Environments Workshop (GCE). 1–8 (IEEE, 2010).

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 12 www.nature.com/scientificreports/ www.nature.com/scientificreports

65. Bouckaert, R. et al. BEAST2: A sofware platform for Bayesian evolutionary analysis. PLoS Computational Biology 10, e1003537 (2014). 66. Leaché, A. D. & Bouckaert, R. R. Species trees and species delimitation with SNAPP: a tutorial and worked example. Workshop on Population and Speciation Genomics, Český Krumlov (2018). 67. Schultheiß, R., Van Bocxlaer, B., Riedel, F., von Rintelen, T. & Albrecht, C. Disjunct distributions of freshwater snails testify to a central role of the Congo system in shaping biogeographical patterns in Africa. BMC Evolutionary Biology 14, 42 (2014). 68. Heath, T. A. Divergence Time Estimation using BEAST v2. Dating Species Divergences with the Fossilized Birth-Death Process (2016). 69. Suyama, Y. & Matsuki, Y. MIG-seq: an efective PCR-based method for genome-wide single-nucleotide polymorphism genotyping using the next-generation sequencing platform. Scientifc Reports 5, 16963 (2015). 70. Takahashi, Y. et al. Lack of genetic variation prevents adaptation at the geographic range margin in a damselfy. Molecular Ecology 25, 4450–4460 (2016). 71. Tamaki, I., Yoichi, W., Matsuki, Y., Suyama, Y. & Mizuno, M. Inconsistency between morphological traits and ancestry of individuals in the hybrid zone between two Rhododendron japonoheptamerum varieties revealed by a genotyping-by-sequencing approach. Tree Genetics & Genomes 13, 4 (2017). 72. Catchen, J., Amores, A., Hohenlohe, P., Cresko, W. & Postlethwait, J. Stacks: building and genotyping loci de novo from short-read sequences. G3: Genes, Genomes, Genetics 1, 171–182 (2011). 73. Falush, D., Stephens, M. & Pritchard, J. K. Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164, 1567–1587 (2003). 74. Evanno, G., Regnaut, S. & Goudet, J. Detecting the number of clusters of individuals using the sofware STRUCTURE: a simulation study. Molecular Ecology 14, 2611–2620 (2005). 75. Earl, D. A. & von Holdt, B. M. STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conservation Genetics Resources 4, 359–361 (2012). 76. Jakobsson, M. & Rosenberg, N. A. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23, 1801–1806 (2007). 77. Rosenberg, N. A. Distruct: a program for the graphical display of population structure. Molecular Ecology Notes 4, 137–138 (2004). 78. Meirmans, P. G. & Van Tienderen, P. H. GENOTYPE and GENODIVE: Two programs for the analysis of genetic diversity of asexual organisms. Molecular Ecology Notes 4, 792–794 (2004). 79. Cornuet, J. M. et al. DIYABC v2.0: a sofware to make Approximate Bayesian Computation inferences about population history using Single Nucleotide Polymorphism, DNA sequence and microsatellite data. Bioinformatics 30, 1187–1189 (2014). 80. Bertorelle, G., Benazzo, A. & Mona, S. ABC as a fexible framework to estimate demography over space and time: some cons, many pros. Molecular Ecology 19, 2609–2625 (2010). 81. Tsuda, Y., Nakao, K., Ide, Y. & Tsumura, Y. Te population demography of Betula maximowicziana, a cool-temperate tree species in Japan, in relation to the last glacial period: its admixture-like genetic structure is the result of simple population splitting not admixing. Molecular Ecology 24, 1403–1418 (2015). 82. Kuhl, F. P. & Giardina, C. R. Elliptic Fourier features of a closed contour. Computer Graphics and Image Processing 18, 236–258 (1982). 83. Iwata, H. & Ukai, Y. SHAPE: A computer program package for quantitative evaluation of biological shapes based on elliptic Fourier descriptors. Journal of Heredity 93, 384–385 (2002). Acknowledgements We thank C. Chen, Z. Don, H. Fukuda, T. Haga, N. Hirano, M. Ishida, Y. Kameda, D. Kanto, Y. Kawachino, K. Kimura, T. Kitano, H. Kubo, K. Mashino, H. Minato, Y. Morii, T. Nagatomi, Okayama Prefectural Nature Conservation Center, J. U. Otani, S. Ozawa, L. Prozorova, T. Sasaki, R. Sato, T. Suenaga, A. Tominaga, S. Uchida, K. Ueda, K. Yamane, D. Yamazaki, and H. Yoshino for sample collection. Tis study was conducted under the support of Zhejiang University, and Vietnam Academy of Science and Technology, and this study complies with CBD regulations/Nagoya protocol (sample number of Vietnam and China are less than 20 individuals). Tis study was partly supported by a research grant from the Research Institute of Marine Invertebrates and JPSP Research Fellow Grant Number 16J04692. Finally we thank two anonymous reviewers and editorial board member for providing the helpful comments and suggestions. Author Contributions T.H. and T.S. conceived the study. T.H., T.S., V.T.D., and B.Y. provided and prepared snail samples. Y.T. and Y.S. conducted MIG-seq method and obtained SNPs data. Y.T. and J.K. provided helpful suggestions of population genetics analyses and performed a part of that. O.M. and S.C. provided benefcial comments regarding whole of the manuscript. T.H. wrote the manuscript with input from all other authors. Additional Information Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-42682-0. Competing Interests: Te authors declare no competing interests. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. Te images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Scientific Reports | (2019)9:6223 | https://doi.org/10.1038/s41598-019-42682-0 13